A Preliminary Study Of MRI-based Radiomics In Evaluating The Early-term Clinical Effects On Advanced Hepatocellular Carcinoma With Hepatic Arterial Infusion Chemotherapy(HAIC)

Aims: To explore the value of radiomics properties extracted from pre-treatment multiparametric magnetic resonance imaging(MRI)in evaluating the early-term clinical effects on patients with advanced hepatocellular carcinoma(Ad-HCC)by using hepatic arterial infusion chemotherapy(HAIC).Methods: This study retrospectively analyzed 85 patients(71 males and 14 females,53.95±11.88 years of age)with advanced liver cancer treated with FOLFOX4-HAIC.The patients were evaluated by the response evaluation criteria in solid tumors(RECIST 1.1)and then divided into the control group and progression group.We extracted the imaging radiomics features of preoperative MRI T2 WI,DWI,and CE-TIWI of patients by using the 3D slicer software.Then,the labels were extracted and selected using statistical methods step by step.A binary multivariate logistic regression radiomics model was constructed by the radiomics features and then tested by using the leave-one-out cross-validation.The receiver operating characteristic curve(ROC)was constructed,and the area under the curve(AUC),sensitivity(Sen),specificity(SPE),and accuracy(ACC)were used to evaluate the predictive values of the radiomics signature.Results: Based on the early-term therapeutic effect,43 cases(50.1%)in the disease control group and 42 cases(49.9%)in the progression group were obtained in the early-term efficacy of FOLFOX-HAIC in the treatment of Ad-HCC.A total of 851 imaging radiomics variables were extracted separately in preoperative single-modality MRI.After dimensionality reduction,4,3,2radiomics features were obtained from T2 WI,DWI,and CE-T1 WI,respectively.The areas under the ROC curves(AUCs)of single-modality T2 WI,DWI,CE-TIWI,combined multimodality MRI radiomics models were 0.807(95% CI 0.714-0.900),0.737(95% CI 0.630-0.844),0.754(95% CI 0.650-0.855),and 0.861(95% CI 0.780-0.942),respectively.The SEN was 0.674,0.860,0.744,and 0.837,respectively.The SPE was 0.786,0.595,0.667 and 0.786,respectively.The ACC was 78.8%,64.7%,70.6% and 81.4%,respectively.There were no significant differences in the prediction efficiency of the model among three different single-modality(P > 0.05).The prediction efficiency of the combined multimodality MRI radiomics model was better than every single-modality model(P < 0.05).Conclusions: The preoperative multimodality MRI radiomics signature offers an outstanding predictive value for the early efficacy of FOLFOX4-HAIC in the early-term treatment of Ad-HCC,and provides certain references for the clinical individualized accurate treatment for tumors of Ad-HCC and risk assessments.